Increased patient volume is associated with improved survival in numerous high-risk surgical and medical conditions. Whether or not a volume-outcome relationship exists within the intensive care unit (ICU) is unknown. The main hypothesis of this proposal is that patients receiving critical care in high volume ICUs will have improved survival and shorter length of stay compared to those in low volume ICUs. To test this hypothesis, the APACHE III database will be used to compare risk-adjusted mortality and ICU length of stay across participating ICUs. Additional multivariate analyses will be performed to adjust for differences in case-mix such as admission source, admission diagnosis, type of hospital, and location of hospital. Hierarchical modeling techniques will be used to account for hospital-level clustering beyond traditional case-mix bias. A greater understanding of the effects of volume on critical care outcome will guide future investigations into the processes of care at high volume centers that may determine the volume-outcome relationship. [unreadable] [unreadable] [unreadable]